NOTE: Many thanks to Akiko Frid for making available these comments on the safety of Monsanto's GMO-maize by the Austrian scientist Dr Alberta Velimirov
Available: http://hejdagmo.files.wordpress.com/2009/12/dr-velimirov-comments-on-austrian-study.pdf

EXTRACT: "But whatever statistical analyses might or might not be used, the data are always the same and open for everybody to see. This study gives the first indication that the stacked event MON810xNK603 led to a reduced reproductive performance as compared to the non-GM line. This result should not be dismissed as 'bad science' but discussed on a scientific level and taken up as incentive to continue investigations in this direction with other GM crops." - Dr Alberta Velimirov

BACKGROUND: In November 2008 the Austrian Ministry of Health publicly released the results of a study it had commissioned to investigate long term reproductive effects in mice following dietary exposure to a genetically modified (GM) stacked maize NK603xMON810.

The study: summary of findings

Mice fed GM insecticide-producing maize over four generations showed a buildup of abnormal structural changes in various organs (liver, spleen, pancreas), major changes in the pattern of gene function in the gut, reflecting disturbances in the chemistry of this organ system (e.g. in cholesterol production, protein production and breakdown), and, most significantly, reduced fertility.

Ref: Biological effects of transgenic maize NK603xMON810 fed in long term reproduction studies in mice. Velimirov A et al. Bundesministerium für Gesundheit, Familie und Jugend Report, Forschungsberichte der Sektion IV Band 3/2008, Austria, 2008. http://bmgfj.cms.apa.at/cms/site/attachments/3/2/9/ CH0810/CMS1226492832306/forschungsbericht_3-2008_letztfassung.pdf

Subsequently two hostile criticisms of the study by ‘regulators’ - EFSA GMO Panel and FSANZ - have been published. Here we present criticisms from each and then Dr. Velimirov's response.

i) From EFSA GMO Panel’s review of the study

The GMO Panel considered this report and came to the following conclusions.  Regarding the RACB [reproductive assessment by continuous
breeding] study, the summary Table 59 contains calculation errors and inconsistencies in the treatment of the data regarding the 3rd and 4th litters. In addition, it seems that the authors have calculated the number of pups at birth per pair and not per delivering pair, which is standard practice. Also, there appears to be methodological deficiencies in the statistical analysis that seriously compromise the interpretation of the data. For the reasons stated above, individual data are required for a proper assessment. In addition, more detailed information regarding the breeding scheme is needed. In particular, it should be clarified whether in the 3rd and 4th pairing the same or different pairs failed to reproduce.  Information regarding the normal variation of the parameters examined in this study for the mouse strain used (historical control data) is required before any conclusion may be drawn on possible alterations in reproductive performance. In addition, further information on
the estrous cycle and histopathological parameters including spermatogenesis, follicle and oocyte counts is essential for assessing the claims of reduced fertility.  The GMO Panel also notes that information on the genetic identity and characteristics of the tested materials is not sufficient.    On the basis of the data presented the GMO Panel is of the opinion that no conclusions can be drawn from the report.

ii) From FSANZ's review of the study

The study report did not indicate how many batches of diet were prepared, the stability of the diets in the form of a meal, or how many samples of corn/diet from each treatment were actually analysed. Based on the presentation of data on the composition of the diet, it appears that only one sample from each diet was analysed. This represents an inadequate sampling regime, given that the authors’ repeatedly stated the importance of examining the potential effect of transgenic feed on physiological and metabolic processes. In addition, the use of a single dietary concentration precludes the establishment of a dose-response relationship, which is a key study design variable supporting the identification of a treatment-related effect, and thus substantially limits the discriminative power of this study.

Summary of Dr Alberta Velimirov's comments:

Design

The RACB design is routinely used for risk assessment processes e.g. by the US National Toxicology Program (NTP) for the last 15 years so far, to establish any possible adverse effects of chemical compounds on the reproductive performance of the test animals. According to the Environmental Protection Agency (EPA), risk assessments of the plant pesticide Bt delta-endotoxin needs no threshold regulation, since it revealed no toxic effects in acute toxicity testing and its exposure level is very low. Therefore no dose-related diets have been tested and the design was adapted to assess whole feed effects.

Results

Generally speaking, feed uptake and parental weight development were comparable throughout the studies.  

In the MGS [multigenerational study] the parameters litter size, litter weight and number of weaned pups were higher in the group fed with the non-GM maize diet in all 4 generations, but not on a significant level. The pup loss was lower in this group (not statistically significant). There was a considerable reduction of females with litters in the 4th generation: out of the 22 females per group only 16 of the non-GM maize group and 14 of the GM maize group had litters, whereas 20 litters were produced in the [also non-GM maize] control group with an Austrian maize variety, constituting the potential base line.  

The same parameters were also measured in the modified RACB. In this case there were statistically significant differences in litter size and weight in the 3rd and 4th litters in favour of the non-GM maize group. But this time pup loss was higher in this group (significant only in the 3rd litters).

Statistical evaluation

It has been suggested, that the statistical methods used to evaluate the results were not state of the art. Especially it was argued that pairs without litters or lost litters were included in the evaluation. Thus it was recommended to strictly relate all data to the number of real litters and not to the number of breeding pairs. Breeding pairs without litters or without live pups at delivery should be excluded from the evaluation. 
But the type of statistical method used clearly depends on the question to be answered. The aim of the project was to compare the reproductive performance between differently fed groups. Thus perished litters as well as pairs without litters but having already delivered litters (not sterile pairs without litters throughout the experiment) have to be included when evaluating group performance.
 
A new statistical approach summarizing the reproductive performance of each female during the test period of the RACB including the co-variate female weight at the beginning of the experiment has been applied and will be evaluated as to its appropriateness (for peer-reviewed publication). The new results show that in the group with the non-GM maize significantly more pups were born in the overall comparison. But the number of pups weaned was only tendentially different in the overall comparison due to the higher pup loss in this group. The litter weights again were significantly higher in the same group.  

But whatever statistical analyses might or might not be used, the data are always the same and open for everybody to see. This study gives the first indication that the stacked event MON810xNK603 led to a reduced reproductive performance as compared to the non-GM line. This result should not be dismissed as 'bad science' but discussed on a scientific level and taken up as incentive to continue investigations in this direction with other GM crops.  

Read the full comments here: http://hejdagmo.files.wordpress.com/2009/12/dr-velimirov-comments-on-austrian-study.pdf